Causal fractal compression of video sequences using matching pursuit

In this article, a new video coding system is proposed scales. The first automatic fractal image compression algorithm which takes advantage of both intrascale and interscale similarities was proposed by Jacquin [4] . In this method, an image is first present in video sequences. Each block in every frame is approxipartitioned into nonoverlapping blocks called range blocks. Each mated with a linear combination of members of an adaptive library range block is then approximated by a linear combination of a of blocks constructed for that block. This library is made up of some fixed constant (dc) block and a transformed version of a selected fixed blocks (e.g., DCT basis blocks) and some adaptive blocks. The single larger block, called a domain block, taken from the same adaptive library blocks are blocks of the same size or larger blocks image. The transformation is a contraction, usually formed by a that are shrunk, taken from the previous frame. For each block in every combination of lowpass filtering, subsampling, and rotation of frame, a rate-distortion optimized fully orthogonal matching pursuit blocks. The selection is usually made by making a library of algorithm is used to determine how many and which ones of the members of the library constructed for that block should be selected transformed domain blocks (which we call adaptive blocks) , and and linearly combined to approximate the block most efficiently in for each range block selecting the library block that gives the terms of bit rate and rms error. Simulation results on sample graybest match to the non-dc component of the range block. scale video sequence ‘‘Miss America’’ suggests that this method has This method was later extended so that it approximated each promising coding performance in terms of PSNR and bit rate, range block with a linear combination of a fixed number of fixed especially compared to other published fractal-based video comblocks, and one single block from the library [5] . In a generalized pression methods. q 1998 John Wiley & Sons, Inc. Int J Imaging Syst coding framework, this method was further generalized [6] to Technol, 9, 305–319, 1998 approximate each range block by a linear combination of any arbitrary number of fixed and adaptive blocks. This is done by